1. Introduction

Humans spend considerable time in front of mirrors, spawning a multibillion-dollar cosmetic industry. Although a lot of species adjust their posture and appearance to suit different situations (e.g., puffing up to increase the appearance of one’s size to predators) there is little evidence to suggest that they are actually aware of what they look like. An exception are our closest living relatives, the great apes, who sometimes show keen interest in their reflections to investigate parts of their bodies that they cannot usually see. When one surreptitiously puts a little paint above the brow of a chimpanzee and then presents a mirror, the ape is likely to examine the mark using its reflection. Members of all great apes have passed this test (Bard, Todd, Bernier, Love, & Leavens, 2006; Gallup, 1970; Lethmate & Dücker, 1973; Posada & Colell, 2007; Povinelli, et al., 1997; Povinelli, Rulf, Landau, & Bierschwale, 1993; Suarez & Gallup, 1981). But there continues to be great disagreement about what this entails about their minds.

2. The Meaning of Self-Recognition

Gordon Gallup, the inventor of this test, advocates a rich interpretation (Gallup, 1970, 1998): Passing the mark test indicates self-awareness - one can become object of one’s own attention. However, with perhaps the possible exception of the most shallow of super models, for most of us the term self-awareness entails a lot more than knowing what you look like. It implies knowledge of your inner world, your likes and dislikes, your personality, skills and attitudes. Perhaps most importantly, it implies some knowledge of where you come from and where you are going. So perhaps Gallup’s interpretation may over estimate what we can reasonably conclude about great ape minds from their passing the test.

On the other hand, there are lean accounts that appear to severely underrate what the task measures. The comparative psychologist Celia Heyes, for instance, argued that all one needs to pass the task is an ability to distinguish feedback from other types of sensory input (Heyes, 1994, 1998). Effectively, any animal that manages to avoid bumping into things, or that avoids biting itself, has demonstrated such ability. Followers of this view therefore dismiss evidence of mirror self-recognition as not indicating anything particularly interesting – while tacitly ignoring the fact that Heyes’ account fails to offer any explanation at all as to why only a few species pass the task and yet many can distinguish feedback from other input in other contexts. For a neutral observer this creates a situation not unlike the climate change debate of recent years: some experts say this, other experts say the opposite, so it is tempting to conclude that we cannot conclude much at all - and it may be best to ignore this all until it is resolved.

As a result scholars and lay people alike may pick whichever interpretation they find more appealing. We can do better than that, and as with understanding climate change, there are some great rewards. One can bring other lines of evidence to bear on the issue. A typical 15-month-old toddler gets very excited when looking in a mirror, but is at a total loss as to where that sticker is that the experimenter put on her fringe. By age two virtually all toddlers pass the test (Amsterdam, 1972; Nielsen & Dissanayake, 2004). My colleagues and I have shown that these children form rapidly updatable expectations of what they look like (Nielsen, Suddendorf, & Slaughter, 2006). In a series of experiments we first demonstrated that young children could equally recognize a mirror image of a mark on their leg as of a mark on their fringe as used in the standard task. In the next experiment, we slipped children into baggy tracksuit pants that we had sewn onto a highchair and then presented them with a mirrored view of their legs. They failed to find stickers on their unfamiliar-looking legs. However, when given 30 seconds exposure to the new pants they were wearing, by removing a tray that blocked the direct view, children passed the task. These results strongly suggest that young children quickly form a mental expectation of what they look like. They rule out theories that place special emphasis on cognition about faces (e.g., Neisser, 1997). The findings do not demand the higher cognitive capacities that Gallup‘s rich account conjectures. But they clearly indicate more than Heyes’ lean account proposes, because children saw the same mirrored feedback of baggy pants in the last two conditions but reacted very differently when they had 30 seconds of prior exposure. Passing the mirror mark test indicates that the subjects have formed a mental expectation of what they look like from the outside.

3. Comparative Conclusions

This strongly suggests that great apes also form such expectations. Chimpanzees develop mirror self-recognition not unlike human children (Bard, et al., 2006). Monkeys, on the other hand, consistently fail the classic task (Anderson & Gallup, 1997; Gallup, Wallnau, & Suarez, 1980; Hauser, Miller, Liu, & Gupta, 2001; Heschel & Burkart, 2006; Roma, et al., 2007). They have also recently failed a leg version of the task (Macellini, Ferrari, Bonini, Fogassi, & Paukner, 2010). High profile cases have been made for success by one elephant (Plotnik, De Waal, & Reiss, 2006), two magpies (Prior, Schwarz, & Gunturkun, 2008) and one dolphin (Reiss & Marino, 2001). However, other studies found negative results (e.g., Povinelli, 1989), so these results clearly require replication. Surprising lessons can be learned from the distribution of the trait among primates.

Emma Collier-Baker and I recently conducted the largest study yet on lesser apes and they all failed the task (Suddendorf & Collier-Baker, 2009). They consistently failed the task in spite of being highly motivated to find the mark. We marked their faces with cake icing to really make sure they want to retrieve the mark when they see it. They greedily scrape it off other body parts they can see directly. We even smeared icing on the mirror itself and the apes would scrape or lick off every last bit of the treat from the mirror and yet ignore the big blob of icing on their own head that was clearly visible in the mirror. These results amount to evidence of absence (rather than the usual common absence of evidence). Great apes and humans do, and lesser apes and monkeys do not recognize themselves, and this has important implication for the evolution of the trait (Suddendorf & Whiten, 2001).

Traits may be shared between species for two profoundly different reasons: analogy and homology (though more complex mechanisms such as horizontal gene transfer also exist). The wings of birds, bats and insects, for example, are independent, analogical solutions to the problem of flight. Homologous traits, on the other hand, are shared because of common descent. To determine whether a trait is shared for homological of analogical reasons, evolutionary biologists compare the number of assumptions each possibility implies about change events that occurred in the past to explain the current distribution of the trait. It is a less parsimonious theory to propose that each bird species independently invented a feathery solution to flying, then to assume that they all inherited the trait by common descent. In our case, if great ape ancestors had evolved a capacity for mirror-self-recognition independently we have to assume that the trait evolved at least on 4 occasions. If great apes have this trait because of common descent, we would have to propose only one such change. It is therefore more parsimonious to conclude that the great ape common ancestor, that in other respects may be quite different from extant apes, acquired this trait before 14 million years ago.

It acquired it before the time that the line leading to orangutans split off and passed it on to all of its descendents. Given that lesser apes fail mirror selfrecognition tasks we can now narrow the emergence of this trait down further to a time after the split from lesser apes. The capacity to rapidly form expectations of what one looks like must hence have evolved between 18 and 14 million years ago (Suddendorf & Collier-Baker, 2009). Phylogenetic reconstruction is quite a powerful way to make an inference about extinct minds even without ever having to lay eyes on a fossil of the creature that first evolved it. Indeed, we do not know what this creature looked like, but it probably knew what it looked like.

This comparative analysis not only informs about the evolution of mind, but also offers entirely new practical opportunities. For example, this analysis can narrow down the search for the neurological and genetic basis of these traits. Because great apes share a homologous capacity for self-recognition, the basis of this trait is to be found among the neuronal and genetic characteristics that are shared by humans and these species. Given that the next closely related species, the lesser apes, do not have the capacity for selfrecognition, we can subtract as not sufficient all those characteristics that they share with great apes and humans. Necessary factors underpinning the trait must hence be found among the genetic and neuronal characteristics that are not shared with lesser apes, but are present in all great apes and humans. For example, recent studies have identified one neuronal characteristic that meets these criteria: so-called Von Economo neurons exist in human and great ape brains, but not in those of lesser apes and monkeys (Nimchinsky, et al., 1999). Such systematic comparative approaches have great potential, over and above the inherently interesting contribution they can make to our understanding of the evolution of our minds. We now know that our capacity to recognize ourselves evolved on Earth in our great ape ancestor between 14 and 18 million years ago.